From model routing and agent reliability to recursive self-improvement, this essay asks how AI capability becomes work that people can verify and trust.
Badges
Badges come from GitHub once the author claims this feed.
Recent posts
every link opens the author’s post-
AI Models Are Getting Smarter. What Comes Next?
-
October 2026 Notes: Learning to Slow Down Amid Fast Feedback
223 October 2026 notes on AI agents, product judgment, attention and slower work, alongside reflections on books, film, relationships and everyday life.
-
Tana Product Teardown: Following a Task from Notes into Meeting Work
Tana and Tana Outliner: follow a task through nodes, meetings and approvals, with design costs, permissions, migration and checks for your first implementation.
-
Ambi Product Teardown: From Remembering to Getting the Work Done
Ambi's listening, memory and background tasks, with privacy boundaries and a testable implementation path.
-
Doneas Product Teardown: Turning a Weekly Plan into Daily Commitments
How Doneas connects weekly plans and daily commitments: its peer circles, rules, pricing, public evidence gaps, and what to test with one small working group.